**** BEGIN LOGGING AT Tue Feb 19 02:59:57 2019 Feb 19 04:40:21 out of curiosity, is it poaaible to flash a device with an spi programmer? Feb 19 04:45:04 that should always be possible (for spi-nor flash), but requires assembling a full flash (including bootloader, bootloader environment, calibration data whatever else is needed) image Feb 19 04:45:57 ah ok. so you can't flash only the bootloader Feb 19 04:46:49 I'm actually not sure if you can only change ranges Feb 19 04:48:28 but as the spi flasher won't know about the partition map, it might actually easier to generate a full flash image Feb 19 07:31:36 mangix: there are many spi programmers, each is somewhat different. Feb 19 07:32:46 i tend to use flashrom, you can set up a map and only flash parts, but that's not to the default, see man page or documentation for details. also get a full copy of the flash several times first (and make sure they match) before proceeding. Feb 19 07:33:27 s/not to the/not the/ Feb 19 07:35:53 often, when you attach the vcc pin to the flash you will power up the CPU and it will start booting, so wait a bit for the spi bus to become quiet. sometimes you can find a system reset pin (often on the jtag header) so you can hold the cpu in reset, which overcomes that problem. Feb 19 07:38:56 e.g. i just flashed a tplink wdr3600 with a pepe2k u-boot and a bigger flash part using this method Feb 19 12:20:55 hmmm any ideas on how to get mips perf stuff working on ath79 archer c7 v2 - I'm *sure* this worked on ar71xx - https://pastebin.com/6BQh9nhM Feb 19 12:41:43 ldir:ping Feb 19 12:47:23 dedeckeh: pong Feb 19 12:49:20 ldir:commit https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=e6d84fa8864e7bbbcec139dd09a2922a06f5b2a0 breaks installation of iproute2 utilities like nstat, ipbridge, etc ... Feb 19 12:49:57 currently they're only installed if the ip-tiny variant is selected while this is not the case for the full-variant Feb 19 12:51:05 oh :-( Feb 19 12:51:07 rework of this commit is required Feb 19 12:51:41 any ideas on how? Feb 19 12:51:57 I'm looking inot it now Feb 19 12:52:00 *into Feb 19 12:52:53 but I want first to understand what Tony Ambardar wants to achieve in the commit Feb 19 12:55:12 so basically he wants to be able to install tc wihtout all the other utilities Feb 19 12:57:15 I don't think that's the intention - the idea was to have a smaller (default) version of libs etc and slightly larger ones that support the BPF stuff. Feb 19 12:57:30 That's as much as I understand it. Feb 19 12:57:59 erm, let me see if I can find the original PRs 'cos there was a lot of discussion. Feb 19 13:00:06 here we go https://github.com/openwrt/openwrt/pull/1627 & 1626 Feb 19 13:02:39 Is it worth re-opening the PR and getting Tony involved? Feb 19 13:05:36 Yes; I wonder why he makes the installation of ip-bridge,ss, nstat,devlink, rdma dependant on the ip-tiny variant Feb 19 13:06:18 ip-full should install them as well Feb 19 13:06:57 similar for the InstallDev step Feb 19 13:07:59 maybe we have to revert the commits for now and re-open the PR as you propose Feb 19 13:08:30 seems like an oversight - I can't speak for Tony - let's revert Feb 19 13:11:38 ok Feb 19 13:15:36 Could you do the revert... I hope just that single commit would do... and I am currently re-opening the PR and letting Tony know. Feb 19 13:20:42 re-opened the PR - do you want me to do the revert? Feb 19 13:21:20 I will test if a single revert is possible Feb 19 13:21:54 thank you Feb 19 13:40:18 ldir: reverting the commit results into dependencies issues for ss on libelf and libintl Feb 19 13:40:35 ldir:we better revert the whole range for now Feb 19 13:45:51 I was afraid that would happen. The ifeq ($(BUILD_VARIANT),tiny) conditional now really doesn't make any sense to me either. Feb 19 13:47:14 I'm testing now if the first two commits can be kept Feb 19 13:47:40 If we just drop the conditionals and in essence do the same things for each variant....does that make more semse? I'm not in a position to test, well not quickly. Feb 19 13:48:23 can give that a try Feb 19 14:04:54 ldir:removing the ifeq ($(BUILD_VARIANT),tiny) conditional results again into the dependency issue of ss on libelf, libintl Feb 19 14:06:32 I'm gonna revert the last four commits of the series for now; let's wait for the feedback on the PR before making further changes Feb 19 14:11:23 ok - I knew this would happen. Feb 19 14:26:38 hi all, is somewhere guide how to generate modified sysupgrade/factory image (change some kconfig option?)? which i currently need after uboot update (changed size occupied by u-boot) Feb 19 14:41:46 dedeckeh: how are you creating the failure? Is this at build time or something else? Feb 19 14:42:58 * ldir 's macbook is burning his knee and eating battery Feb 19 14:43:13 better than the otherway around Feb 19 14:44:05 yeah lithium ion battery fires are a right bastard to put out ;-) Feb 19 14:45:35 ldir: at build time Feb 19 14:46:39 dedeckeh: ok good - so if I can put the fire out I should be able to replicate - why didn't the buildbots fall over this? Or maybe they did? Feb 19 14:46:44 * ldir doesn't dare look Feb 19 14:47:57 ldir:compile issue is hit if the ifeq ($BUILD_VARIANT),tiny) conditional is removed and ss for example is enabled Feb 19 14:48:24 I gues that was the reason Tnoy put that in place but obviously this is wrong Feb 19 14:48:32 *Tony Feb 19 14:48:35 * ldir opens one eye slowly on buildbots Feb 19 14:49:52 did you break it again, yes? :P Feb 19 14:53:24 stintel: yes - again, again, again Feb 19 14:55:26 booooo! :D Feb 19 14:57:59 lol - indeed boooo. I'm thinking about that. But first it's 1500 and the alarm spaniel has gone off demanding its walk. Feb 19 14:59:24 it *has* taken a number of days to reveal that it's gone wrong. ugh - bggr bggr bggr bggr Feb 19 15:00:28 hehe, I've been writing a nagios check for HPE Aruba gear, and now it turns out that CentOS/EPEL pysnmp is broken Feb 19 15:00:30 fml Feb 19 15:02:20 can i trigger a firmware upgrade through uci or some other CLI command for the 'recovery' instead of the 'sysupgrade' image? Feb 19 15:02:57 no, wait, the sysupgrade image contains all the files, right? Feb 19 15:24:14 jow: ping you there? Feb 19 17:19:18 dedeckeh: ping Feb 19 17:21:26 dedeckeh: In a 'pre-revert' state I cannot replicate the issue. Feb 19 17:30:39 ldir:issue being ss,nstat, ip-bridge not being installed in case iā»full is selected ? Feb 19 17:30:47 ip-full Feb 19 17:32:06 so I have ip-full selected & nstat and I build but don't (yet) see an error Feb 19 17:32:25 ypu don' see an error; just they're not installed Feb 19 17:34:19 ahh, ok! At least now I understand the failure mode :-) Feb 19 17:34:30 I tried to remove the ifeq ($(BUILD_VARIANT),tiny) condition to fix the above issue Feb 19 17:34:47 Do they even get built (ldir waits for his macbook to stop burning a hole in his leg) Feb 19 17:35:08 but that resulted in a dependency issue of ss on libelf Feb 19 17:35:45 yes they get built but are not installed Feb 19 17:36:24 ok, all good info for Tony when he appears. Feb 19 17:37:28 indeed Feb 19 17:39:56 AFAIUI tc is the only thing that should be interested in libelf - for loading of BPF programs stored in ELFs Feb 19 19:31:01 Any suggestions for where / when to look for discussions around why extraction of device/board-specific ath10k ART is done with hotplug scripts? Feb 19 19:31:35 Seems too late relative to driver start, at least with the device I'm looking at Feb 19 19:34:55 meep Feb 19 19:36:20 * blogic notes that cisco anyconnect is an insult ot humanity Feb 19 19:36:40 dwmw2_gone: ^^ why is there no MFA support in openconnect ;) Feb 19 19:37:00 blogic: well, they have put ASA out to the world, that's even worse. Feb 19 19:37:08 the evil offspring of the demonic PIX Feb 19 19:37:16 do i want to know what that is ? Feb 19 19:37:32 blogic: it's their low-end enterprise firewall. Feb 19 19:37:37 oh dear Feb 19 19:38:00 "Cisco ASA 5500-X with FirePOWER Services" Imagine the meeting where they came up with that? Feb 19 19:38:25 they were on crack Feb 19 19:38:30 or worse Feb 19 19:38:31 Larry Demar pinball fans, perhaps? Feb 19 19:38:46 cracktasy, gets you high by looking at it Feb 19 19:40:58 woah, it's a blogic Feb 19 19:41:10 blogic: Did you get any luck getting that datasheet from MTK? Feb 19 19:41:48 pfff Feb 19 19:41:59 i am out of the loop since months Feb 19 19:42:07 just getting back into real life Feb 19 19:42:14 real life hit me Feb 19 19:42:33 and then i had a car crash 2 weeks ago Feb 19 19:42:36 :{ Feb 19 19:42:43 Any real damage? Feb 19 19:42:51 yeah car full of 6 year old giurls on the way to the zoo :-) Feb 19 19:43:03 only metal, 12k of it ... Feb 19 19:43:11 Metal can be replaced Feb 19 19:43:15 my car was a write-off Feb 19 19:43:38 picked up the new car today after what feels like 100 calls with insurances and banks Feb 19 19:43:53 BUT, i now have a polo with 122HP Feb 19 19:43:58 hrr hrr Feb 19 19:44:03 You expected them to give up their easily earned profits easily? Feb 19 19:45:11 not really Feb 19 19:49:24 blogic: So someone found the magic bit in the MT7688 datasheet, although I still have no idea what the adjusted pinout is for the SD interface. Feb 19 19:53:41 Hopefully get back in gear soon and see if I'm smart enough to make all that work properly Feb 19 19:54:12 * Monkeh glares at the bricked Ubiquiti on his desk Feb 19 19:55:33 Monkeh: i have a box for the bricks Feb 19 19:55:45 Eh, I'm hoping to find time to fix that, too Feb 19 19:55:58 It's only an old one Feb 19 19:56:20 Needs moving to ath79, small mtdsplit issue. Feb 19 19:56:29 i nuked a unit yesterday Feb 19 19:56:46 early production unit for a customer, there were only 3 of them :-) Feb 19 19:57:01 great start into the week Feb 19 19:57:11 and it has nand flash so no way to recover it Feb 19 19:57:42 And by 'small issue' I mean 'the crazy fools put the kernels in jffs2' Feb 19 19:58:18 Ahh, the advantages of NAND.. :) Feb 19 19:58:51 Monkeh: that's still a bit better than putting the kernels raw on nand Feb 19 19:58:54 It's so nice with spi-nor, I can just pop the chip off and reflash it. Feb 19 19:58:59 KanjiMonster: But it's not NAND Feb 19 19:59:04 So it's just more annoyance for me Feb 19 19:59:11 ah, I see Feb 19 19:59:35 I just haven't had a chance to sit down and figure out WTF to do with it Feb 19 19:59:43 Too many other bits on the go Feb 19 20:04:00 blogic: sorry to hear Feb 19 20:14:31 9 Feb 19 20:29:25 I was right - perf does work on ar71xx on archer c7 v2 - but not on ath79 https://pastebin.com/6BQh9nhM Feb 19 20:30:12 ldir: perf irq setup is prolly missing Feb 19 20:45:56 blogic: is that a DTS thing? I looked in target/linux/ar71xx/files/arch/* for clues but failed Feb 19 20:46:39 "perf" ?? Feb 19 20:48:15 cpu performance counters so you can get a flavour of where the system is spending cpu time. Feb 19 20:55:07 Ah, thanks! **** ENDING LOGGING AT Wed Feb 20 02:59:57 2019